[tomhow]

It's standard practice to take those kinds of numbers out of title, because they make the title more baity, and often cause much of the discussion to focus on debate about how accurate/normal the figure is. It's sufficient for the title to say "faster" then let the article demonstrate how much faster it can be in different scenarios.

[nine_k]

But there is a qualitative difference between 5% faster and 95% faster: the latter usually meaning a serious rework, and the former being a small incremental improvement.

I'd be okay with replacing "95% faster" with "several times faster" to still convey the point.

[tomhow]

It's not about the size of the number or improvement; we do the same thing when the number is "10,000%", which is not unusual in the titles we see here.

The problem with these kinds of titles – and this is no comment on this particular article (I haven't checked, because it's irrelevant) – is that sometimes writers will put a figure in the title that was achieved in a one-off result under very specific/unusual conditions, whereas the realistic improvement under more normal conditions is like 20% or 50% – still great, just not what the title claimed.

Then when that happens, the discussion becomes dominated by comments pointing that out and debating the validity of the tests and results – even if the article does a good job of revealing those details.

We've found we can reduce that effect by taking the numbers out of the title altogether.